SEMANTIC INTELLIGENCE FOR KNOWLEDGE-BASED COMPLIANCE CHECKING OF UNDERGROUND UTILITIES
Underground utilities must comply with the requirements stipulated in utility regulations to ensure their structural integrity and avoid interferences and disruptions of utility services. Noncompliance with the regulations could cause disastrous consequences such as pipeline explosion and pipeline contamination that can lead to hundreds of deaths and huge financial loss. However, the current practice of utility compliance checking relies on manual efforts to examine lengthy textual regulations, interpret them subjectively, and check against massive and heterogeneous utility data. It is time-consuming, costly, and error prone. There remains a critical need for an effective mechanism to help identify the regulatory non-compliances in new utility designs or existing pipelines to limit possible negative impacts. Motivated by this critical need, this research aims to create an intelligent, knowledge-based method to automate the compliance checking for underground utilities.
The overarching goal is to build semantic intelligence to enable knowledge-based, automated compliance checking of underground utilities by integrating semantic web technologies, natural language processing (NLP), and domain ontologies. Three specific objectives are: (1) designing an ontology-based framework for integrating massive and heterogeneous utility data for automated compliance checking, (2) creating a semi-automated method for utility ontology development, and (3) devising a semantic NLP approach for interpreting textual utility regulations. Objective 1 establishes the knowledge-based skeleton for utility compliance checking. Objectives 2 and 3 build semantic intelligence into the framework resulted from Objective 1 for improved performance in utility compliance checking.
Utility compliance checking is the action that examines geospatial data of utilities and their surroundings against textual utility regulations. The integration of heterogeneous geospatial data of utilities as well as textual data remains a big challenge. Objective 1 is dedicated to addressing this challenge. An ontology-based framework has been designed to integrate heterogeneous data and automate compliance checking through semantic, logic, and spatial reasoning. The framework consists of three key components: (1) four interlinked ontologies that provide the semantic schema to represent heterogeneous data, (2) two data convertors to transform data from proprietary formats into a common and interoperable format, and (3) a reasoning mechanism with spatial extensions for detecting non-compliances. The ontology-based framework was tested on a sample utility database, and the results proved its effectiveness.
Two supplementary methods were devised to build the semantic intelligence in the ontology-based framework. The first one is a novel method that integrates the top-down strategy and NLP to address two semantic limitations in existing ontologies for utilities: lack of compatibility with existing utility modeling initiatives and relatively small vocabulary sizes. Specifically, a base ontology is first developed by abstracting the modeling information in CityGML Utility Network ADE through a series of semantic mappings. Then, a novel integrated NLP approach is devised to automatically learn the semantics from domain glossaries. Finally, the semantics learned from the glossaries are incorporated into the base ontology to result in a domain ontology for utility infrastructure. For case demonstration, a glossary of water terms was learned to enrich the base ontology (formalized from the ADE) and the resulting ontology was evaluated to be an accurate, sufficient, and shared conceptualization of the domain.
The second one is an ontology- and rule-based NLP approach for automated interpretation of textual regulations on utilities. The approach integrates ontologies to capture both domain and spatial semantics from utility regulations that contain a variety of technical jargons/terms and spatial constraints regarding the location and clearance of utility infrastructure. The semantics are then encoded into pattern-matching rules for extracting the requirements from the regulations. An ontology- and deontic logic-based mechanism have also been integrated to facilitate the semantic and logic-based formalization of utility-specific regulatory knowledge. The proposed approach was tested in interpreting the spatial configuration-related requirements in utility accommodation policies, and results proved it to be an effective means for interpreting utility regulations to ensure the compliance of underground utilities.
The main outcome of this research is a novel knowledge-based computational platform with semantic intelligence for regulatory compliance checking of underground utilities, which is also the primary contribution of this research. The knowledge-based computational platform provides a declarative way rather than the otherwise procedural/hard-coding implementation approach to automate the overall process of utility compliance checking, which is expected to replace the conventional costly and time-consuming skill-based practice. Utilizing this computational platform for utility compliance checking will help eliminate non-compliant utility designs at the very early stage and identify non-compliances in existing utility records for timely correction, thus leading to enhanced safety and sustainability of the massive utility infrastructure in the U.S.